Index  | Recent Threads  | Unanswered Threads  | Who's Active  | Guidelines  | Search
 

Quick Go »
No member browsing this thread
Thread Status: Active
Total posts in this thread: 5
[ Jump to Last Post ]
Post new Thread
Author
Previous Thread This topic has been viewed 1144 times and has 4 replies Next Thread
nickoli
Senior Cruncher
USA
Joined: Nov 28, 2005
Post Count: 167
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
devilish Consensus on HT and BOINC

Hi All,

Sorry for another Hyperthreading thread. I've read quite a few and haven't observed a consensus on the issue.

So...

I have a rig running Ubuntu, with (2) Xeons @ 3.6GHz, with 2GB RAM, for a total of 2 virtual and two real cores. I currently have it set to use 50% of multiprocessors... Is this what others find most productive with Xeon procs?

Thanks in advance; I'm sorry for just another HYPERTHREADING-Thread!

biggrin
----------------------------------------

[Jul 13, 2013 5:26:29 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Consensus on HT and BOINC

Hello Nickoli,
The new Intel cores run about 25$ more total throughput while hyperthreading.But the problem is that you only have 2GB which has to run the OS as well. If you are not running CEP2, just FAAH, you might try 75% or even 100%. If you had 3 or 4 GB, you could do best at 100% but with your memory you might be at the optimum. You could try experimenting while checking memory usage, speed, etc.

Lawrence
[Jul 13, 2013 5:54:38 PM]   Link   Report threatening or abusive post: please login first  Go to top 
nickoli
Senior Cruncher
USA
Joined: Nov 28, 2005
Post Count: 167
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: Consensus on HT and BOINC

Hi Lawrence,

So, I'm running a test. Changed my percent of processors used from 50% to 75% — new WU downloaded. The new WU is FAAH, the other two are FAAH and CEP. Memory usage went from 23%~ to 27.5%.

I am thinking about going to 100%, but I think that may reduce my efficiency and make WUs run longer.

Thoughts?
----------------------------------------

[Jul 14, 2013 12:35:23 AM]   Link   Report threatening or abusive post: please login first  Go to top 
KWSN - A Shrubbery
Master Cruncher
Joined: Jan 8, 2006
Post Count: 1585
Status: Offline
Reply to this Post  Reply with Quote 
Re: Consensus on HT and BOINC

It will make WU run longer but that's not the point. In many situations HT will increase your overall points per hour even though the units have longer run times.

What Lawrence suggested is the only way you can be certain if HT is the best answer for your particular system and circumstances. Run at 50% on a single science for a set period of time, say a week, then do the same test at 100% (or whatever percentage) and compare the long-term averages. Anything less will be too susceptible to statistical flukes and will not give you a true answer.
----------------------------------------

Distributed computing volunteer since September 27, 2000
[Jul 14, 2013 2:08:58 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Former Member
Cruncher
Joined: May 22, 2018
Post Count: 0
Status: Offline
Reply to this Post  Reply with Quote 
Re: Consensus on HT and BOINC

Hi Nickoli,
Your memory figures look good. You can certainly run 1 CEP2 and 2 FAAH threads at 75% or 4 FAAH at 100%. Personally I would run 100% and run a CEP2 thread and 3 FAAH after seeing those memory figures. CEP2 never hets a fair share.

Lawrence
[Jul 14, 2013 4:23:30 AM]   Link   Report threatening or abusive post: please login first  Go to top 
[ Jump to Last Post ]
Post new Thread